Sunteți pe pagina 1din 11

Ocean Engineering 80 (2014) 25–35

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Numerical investigation of a fleet of towed AUVs


Pareecha Rattanasiri n, Philip A. Wilson, Alexander B. Phillips
Fluid Structure Interactions Group, Engineering and the Environment, University of Southampton, Southampton SO17 1BJ, United Kingdom

art ic l e i nf o a b s t r a c t

Article history: This paper investigates the influence on the fleet of the drag of multiple towed prolate spheroids to
Received 26 April 2013 determine the hydrodynamic effect of the viscous interaction between hulls and to study the influence of
Accepted 4 February 2014 the configuration's shape of multiple hulls in the vee and echelon formations. A series of CFD RANS-SST
simulations has been performed at the Reynolds Number 3.2  106 by a commercial code ANSYS CFX 12.1.
Keywords: Mesh convergence is tested and then validated with experimental and empirical results. The drag of each
Co-operative AUVs spheroid is compared against the benchmark drag of a single hull. The results show that the spacing
Fleet of AUVs between two hulls determines the individual drag and combined drag. The dominant spacing has been
Towed model classified into seven zones based on the drag characteristic of twin towed models. Regions are
Vee formation
characterised to parallel, echelon, no gain, push, drafting, low interaction, and no interaction. Both the
Echelon formation
multi-vehicle vee and echelon configurations show limited influence against that of the entire fleet's
CFD RANS-SST simulation
energy budget. For an individual spheroid where a lower propulsion cost is required, then the use of
three/four in vee or echelon formation should be considered. Based on this numerical information,
operators can determine the optimal fleet configuration based on energy considerations.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction endurance. The total power consumed by an AUV can be split into
propulsion power and the hotel load required to power non-
Autonomous Underwater Vehicles (AUVs) are self-propelled propulsion related electronics and sensors (Phillips et al., 2012).
robots which perform missions without requiring external power- This study only considers propulsion power.
ing or an umbilical control. Current AUVs have the capability to For an individual AUV the propulsion power may be reduced by
perform missions such as pipeline inspection (Labbe et al., 2004; obtaining the optimum hydrodynamic design of such as hull,
Weiss et al., 2003), mine-sweeping (Edwards et al., 2004), and propeller and surface control (Bellingham and Willcox, 1996;
oceanographic exploration (Botelho et al., 2005; Martins et al., Bradley, 1992; Bradley et al., 2001; Dalton and Zedan, 1980;
2003). Proposed missions for the next generation of AUVs require Huggins and Packwood, 1994; Jagadeesh and Murali, 2005, 2006;
them to acquire both high quantity and quality temporal and Jagadeesh et al., 2009; Kinsey, 1998; Parsons, 1972; Parsons et al.,
spatial data. This enhanced performance may be achieved by 1974; Phillips, 2009; Sarkar et al., 1997a; Stevenson et al., 2007).
either homogeneous or heterogeneous fleets of vehicles. By carry- For a fleet of multiple AUVs, minimising the energy consumption
ing different sensors, the surveying performance of multiple small may be targeted for both individuals and the entire fleet. Previous
vehicles may have a performance equal to or exceeding that of a experimental studies of flow past bodies in close proximity such as
single larger vehicle, with improved levels of robustness and two circular cylinders, two slender bodies, a car with trailer and
redundancy for a specific task. This has led to studies of man- cars in convoy demonstrate the effect of relative distance between
oeuvring and path planning of fleets by Aguiar et al. (2011), Bean the bodies on the fleet drag as a whole (Hoerner, 1965; Hucho and
et al. (2007), Burger et al. (2009), Cui et al. (2009, 2010), Reeder Ahmed, 1998). When slipstreaming in cycling, the trailing riders
et al. (2004), and Vanni (2007). experience a 38% reduction in wind resistance (Kyle, 1979).
Since the range of an AUV is dictated by its finite energy source, Drafting in swimming, the relative drag of the second swimmer
minimising the energy consumption is required to maximise in line is approximately 56% and 84% lower than the leading
swimmer, when the distance between the swimmers is 0.5 m and
6.0 m, respectively (Silva et al., 2008). Observation of animal
n
Correspondence to: Fluid Structure Interactions Group (Room 28/1021), motion such as fish in schools, dolphins in pods or birds in flocks
Engineering and the Environment, Highfield Campus, University of Southampton,
suggests some energy benefit may be obtained by certain fleet
Southampton SO17 1BJ, United Kingdom. Tel.: þ 44 23 8059 5097.
E-mail addresses: pj506@soton.ac.uk (P. Rattanasiri), configurations (Alexander, 2004; Andersson and Wallander, 2003;
philip.wilson@soton.ac.uk (P.A. Wilson), abp@soton.ac.uk (A.B. Phillips). Hanrahan and Juanes, 2001; Partridge et al., 1983; Weihs, 2004).

http://dx.doi.org/10.1016/j.oceaneng.2014.02.001
0029-8018 & 2014 Elsevier Ltd. All rights reserved.
26 P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35

Nomenclature dm maximum diameter of the body of revolution (m)


D longitudinal offset (m)
1þ k the form factor of a single spheroid D/L non-dimensional longitudinal offset
1þ βk the form factor of twin spheroids dh maximum hull diameter (m)
Aw wetted surface area (m2) S/L transverse separation
B1 leading spheroid L length of the body from nose to tail (m)
B2, B3, and B4 following spheroid Re length Reynolds number, VL/ν
%B1, %B2, %B3, and %B4 individual drag of B1, B2, B3, and B4 V vehicle speed (m/s)
referred to a single hull drag Ui Cartesian mean velocity components (Ux, Uy, Uz)
%CB combined drag refer to sum of two single hull drags xi represents Cartesian co-ordinates (X, Y, Z) (m)
CD total drag coefficient, Total drag/(0.5ρV2Aw) ν fluid kinematic viscosity, μ/ρ (m2/s)
CD(B1) and CD(B2) drag coefficient of B1 and B2 in fleet ρ fluid density (kg/m3)
CD(B3) and CD(B4) drag coefficient of B3 and B4 in fleet μ fluid dynamic viscosity (kg/m s)
CD(s) drag coefficient of a single hull k fluid turbulent kinetic energy (m2/s2)
CF skin friction drag coefficient, Skin friction drag/ ε rate of dissipation of turbulent energy (m2 s  3)
(0.5ρV2Aw) ϕ the advection scheme
(CF)1957 ITTC'57 skin friction line Ru the vector from the upwind node to the
CPx pressure drag coefficient, Pressure drag/(0.5ρV2Aw) integration point
CSF coefficient of side-force

For mother and calf dolphins, the calf derives a 60% energy benefit
by taking up a hydrodynamically advantageous position in close
proximity to its mother (Weihs, 2004). This assumption has agreed
with the experimental results of twin prolate spheroids in the
wind tunnel (Weihs et al., 2007). Recently, Husaini et al. (2009)
numerically illustrated the influence of distance among multiple
towed torpedo shape AUVs to their individual drags. These results
suggest the energy benefit by using a fleet of AUVs may be
obtained at the optimal configuration with the optimal distance
and lead to the underlying questions of:

 Does a fleet configuration provide energy benefits for just an


individual AUV or the whole fleet?
 What is the optimal configuration and optimal distances of
the fleet?

The purpose of this paper is to provide guidance for operators


on suitable spacing for multiple vehicles' configurations. Based on
this numerical information, operators can determine the potential
configuration with the optimal distances based on the energy
consideration. To achieve this aim, two main hydrodynamic
processes of twin towed hulls: the body-to-body interference (or
viscous interaction) and the increase of drag of following hulls due
to the wake of a leading hull must be numerically investigated,
firstly, the effect of various configurations to the individual drag,
the fleet drag and the interference drag of a series of two prolate Fig. 1. Physical flow problem: (a) flow passes a single hull, (b) flow passes twin
spheroids. Secondly, the influence of the number of spheroids in hulls in parallel configuration, and (c) flow passes twin hulls in drafting
configuration.
the fleet including the shape of the configuration must be
numerically investigated.
to the viscous drag (Cv) and can be estimated as

2. Theoretical approach C D  C v ¼ ð1 þ kÞC F ¼ ðTotal dragÞ=ð0:5ρV 2 Aw Þ ð2Þ

2.1. Drag estimation where ρ is the fluid density, Aw is the wetted surface area and V is
the fluid velocity and a form factor (1 þk) was introduced to
The flow pass a single bare hull, twin hulls in parallel config- estimate the ratio of total viscous drag (Cv) to the skin friction drag
uration and twin hulls in drafting configuration being considered (CF). CF may be estimated by the ITTC'57 correlation line (ITTC,
here are sketched in Fig. 1a, b and c, respectively. To estimate total 1957):
drag coefficient for fluid flow pass a single hull can be obtained by
ðC F Þ1957 ¼ 0:075=ðlog 10 ðReÞ  2Þ2 ð3Þ
C D ¼ C wave þ C v ð1Þ
Assuming the AUV is fully submerged in deep water, there will be For streamlined shapes, Hoerner (1965) suggests an estimated
no wave resistance (Cwave E0), the total drag is therefore only due form factor in terms of the body length (L) and the maximum body
P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35 27

diameter (dm) as where ϕip is the value at the integration point, ϕup the value at the
3=2 3 upwind node and Ru the vector from the upwind node to the
ð1 þ kÞ ¼ 1 þ 1:5ðdm =LÞ þ 7ðdm =LÞ ð4Þ
integration point. The model reverts to first order when b¼0 and
In the case of parallel twin hulls in close proximity, the is a second-order upwind biased scheme for b¼1. The high-
conventional form factor for a single hull cannot establish an resolution scheme calculates b using a similar approach to that
accurate prediction due to the accelerated flow velocity between of Brath and Jesperson (1989), which aims to maintain b locally to
twin hulls, as shown in Fig. 1b. This is called a viscous interaction be as close to one as possible without introducing local oscilla-
effect which results in an increase of the drag of both hulls. In case tions. Collocated (nonstaggered) grids are used for all transport
of drafting twin hulls in close proximity, the pressure distribution equations, and pressure velocity coupling is achieved using an
around the tail of the leading hull interacts with that around the interpolation scheme based on that proposed by Rhie and Chow
stagnation point of following hull (as shown in Fig. 1c). The (1982). Gradients are computed at integration points using tri-
pressure distribution of each hull is then changed which cause linear shape functions defined in ANSYS (2010) CFX. The linear set
the individual drag to increase/decrease. By combining a viscous of equations that arise by applying the finite volume method to all
resistance factor β, the estimation of a more accurate form factor elements in the domain are discrete conservation equations. The
(1 þβk) for twin hulls can be obtained (Molland and Utama, 2002), system of equations is solved using a coupled solver and a
hence the total drag of twin hulls can be estimated as multigrid approach.
C D ¼ ð1 þ βkÞC F ð5Þ The shear stress transport (SST) turbulence closure model
(Menter, 1994) which blends k  ε and k  ω was selected to close
To predict the hydrodynamic forces acting on an AUV's hull, a the mathematical model for this study. Previous investigations
steady-state Reynolds Averaged Navier Stokes (RANS) simulation have shown that it is better able to replicate the flow around the
has proved to provide reasonably accurate results when compared ship and submarine hull form than either k  ε or k  ω model,
against the experimental results (Jagadeesh et al., 2009; Karim notably with a moderate accuracy (Larsson and Baba, 1996;
et al., 2009; Phillips et al., 2008, 2007, 2010b; Sarkar et al., 1997b). Phillips et al., 2010a).
However, to obtain a high fidelity result needs an appropriate
mesh resolution to capture the effect of the boundary layer, body-
to-body interaction and the wake behind the body.
3. Case study

2.2. Reynolds Averaged Navier Stokes (RANS) 3.1. Base experiment

By assuming the flow is incompressible, the continuity equa- Molland and Utama (1997) performed a series of experiment
tion becomes on twin prolate spheroids to characterise the side-force and
∂U i yawing moment interactions. Tests were carried out in the 70  50
¼0 ð6Þ
∂xi (2.20 m  1.57 m) low speed wind tunnel at the University of
Southampton. The overall length (L) of each model was
The momentum equation can be written as
! ( !) 1200 mm with maximum diameter (dh) of 200 mm and a surface
∂U i ∂U i U j ∂P ∂ ∂U i ∂U j ∂ui'uj' area (Aw) of 0.601 m2. The top spheroid was placed at the middle
ρ þ ¼ þ μ þ ρ þf i ð7Þ
∂t ∂xj ∂xi ∂xj ∂xj ∂xi ∂xj breadth and 1.07 m height from the floor. It was fitted to the
overhead wind tunnel dynamometer for measuring the total drag
where the tensor xi represents Cartesian co-ordinates (X, Y, Z) and and side-force. The lower spheroid (B2) was placed at the
Ui are the Cartesian mean velocity components (Ux, Uy, Uz). The transverse separation (S/L) of 0.27, 0.37 and 0.47 away from the
Reynolds stress tensor ðρu0i u0j Þ is represented the turbulence closure centerline of B1. The noses of both spheroids are aligned with zero
and f i is the external force. longitudinal offset (D/L ¼0) as shown in Fig. 2.
The RANS equations are implemented in the commercial CFD
code ANSYS (2010) CFX. The governing equations are discretised
using the finite volume method. The high-resolution advection 3.2. Present study
scheme was applied for the results presented which varies
between first- and second order accuracy depending on spatial To verify the numerical strategies and results, Molland and
gradient. For a scalar quantity ϕ the advection scheme is written in Utahan's (1997) experiment of twin towed bare prolate spheroids
the form of with transverse separations, where D/L ¼0 and S/L¼ 0.17, 0.27, 0.37
and 0.47, are selected. Where D is the longitudinal distance
ϕip ¼ ϕup þ b∇ϕ URu ð8Þ between each nose of bodies, S is the transverse distance between

Fig. 2. Set up of twin prolate spheroids in the 14 m long, 70  50 wind tunnel.


source: Figure adapted from Molland and Utama (1997).
28 P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35

Fig. 3. Fleets of prolate spheroids with the body length L where S/L is the transverse separation and D/L is the longitudinal offset: (Left) twin hulls configuration, (Middle) vee
formation and (Right) echelon formation.

centre lines of each bodies and L is the body length. As shown in


Fig. 3, the cases have been extended to study the following:

 Firstly, the effect of longitudinal spacing on the drag. The


distances between two hulls are considered at various combi-
nations between 0 oD/L r1.77 and 0.27 oS/Lr0.47. The result
is shown in Section 4.
 Secondly, the effect of the configuration's shape of multiple
hulls on the drag. The symmetry configuration and the echelon
configuration are considered. The result is shown in Section 5.
The case details are as follows:
 Fleets of three spheroids in vee formation for S/L¼ 0.17, 0.27
and 0.37 and D/L¼ 0.27, 0.57, 0.87 and 1.17.
 Fleets of the three/four spheroids in echelon for S/L ¼0.17
and D/L ¼0.27 and 0.57.

3.3. Numerical settings

To replicate the experiment of Molland and Utama (1997), the


dimension of fluid domain is modelled at 1.4L  12L  1.8L. Free
slip wall conditions (ANSYS, 2010) are used for the roof, floor and
walls. The air inlet velocity (V) is set at 40 m/s related to a length
Reynolds number (Re) 3.2  106 (typical AUV operates at Re
between 105 and 107) with the zero relative pressure outlet
boundary condition. The air density (ρair) and the air kinematic
viscosity (υair) at room temperature are 1.185 kg/m3 and
1.545  10  5 m2/s, respectively. Both hulls are modelled by using
no slip wall condition (ANSYS, 2010).

3.4. Mesh strategy

The sample of meshes cut in the ZX plane and the YZ plane is


shown in Fig. 4a and b, respectively. The model domain, boundary
Fig. 4. Mesh cut around a pair of spheroids for S/L ¼0.27 and D/L ¼0 for medium
condition and mesh strategies used in this simulation are illu-
mesh (case f). (a) ZX plane at the centre line at Y ¼ 0 with the fluid flow from left to
strated in Fig. 5. The results of drag components; the total drag right and (b) YZ plane at X ¼ 0.6 m from the noses.
coefficient (CD), the pressure drag coefficient (CPx) and the skin
friction drag coefficient (CF) and form factor (1 þ k), for each cases
are shown in Table 1. A series of eleven meshes ranging from 1.2 to (Molland and Utama, 2002) and empirical results (Hoerner, 1965)
22.7 million elements were made for an example case, where the are as follow.
1.2, 8.6 and 22.7 million element is considered as the coarse mesh,
the medium mesh and the fine mesh, respectively. With 2.0% 3.5.1. Pressure distribution
difference of total drag between the medium mesh (case f) and the The distribution of the pressure coefficient, CP ¼(P  P1)/
fine mesh (case j), the computational cost of using the medium (0.5ρU2), is shown in Fig. 6a–c. Where, X/L is a position on the
mesh is approximately 8 h which is 10 times lower than that of longitudinal axis of the body. A good agreement is observed
fine mesh when using the 64 bit operating computational system, between the numerical trend lines and the experimental results
2.53 GHz, two Core Processors with 12 GB RAM. Thus the mesh for the three separated separations.
parameters of global mesh size and the local refinement case f is
selected as appropriate. The computational parameters are pro-
vided in Table 2. 3.5.2. Side-force coefficient
The side-force is defined as an attraction force between the
3.5. Results and validation hulls (Fz). The side-force coefficient (CSF) is defined by CSF ¼ Fz/
(0.5ρU2). In Fig. 7, the numerical results of CSF of B1 show excellent
The twin towed bare hulls are compared against the experi- agreement with the experimental results and show that previous
mental results (Molland and Utama, 1997), the numerical results numerical demonstrations were deficit because of the lack of high
P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35 29

Fig. 5. (Top) Boundary condition and mesh refinement around a pair of prolate spheroids and (Bottom) eleven mesh strategies.

Table 1 Table 2
The result of drag components; the total drag coefficient (CD), pressure drag Computational parameters.
coefficient (CPx), skin friction drag coefficient (CF) and form factor (1 þ k), for each
mesh strategies. Parameters Setting

Case CD  103 CPx  103 CF  103 (1 þ k) Mesh type Unstructured with local refinement around
spheroids and in wake regions
a 4.325 0.635 3.718 1.163 yþ Average 30
b 4.307 0.625 3.714 1.160 No. of elements 8–15 M with 15–20 prism layers in the boundary layer
c 4.260 0.571 3.717 1.146 Turbulence model Shear stress transport
d 4.231 0.550 3.713 1.140 Inlet turbulent 1%
e 4.180 0.493 3.715 1.125 intensity
f 4.168 0.481 3.711 1.123 Wall modelling Automatic wall function (ANSYS, 2010)
g 4.154 0.449 3.724 1.115 Spatial discretisation High resolution (ANSYS, 2010)
h 4.139 0.388 3.751 1.103 Timescale control Auto timescale (ANSYS, 2010)
i 4.124 0.379 3.745 1.101 Convergence criteria RMS residualo 10  6
j 4.086 0.355 3.730 1.095 Computing 3 Linux cluster
k 4.076 0.344 3.730 1.093 Run type Parallel (12 partitions run on 4  dual core
nodes, each with 2 GB RAM)
Simulation time 2.0–2.5 Wall clock hours (medium mesh)
performance computers at that time. The rapidly decrease of the
side-force ratio with an increase of separation, is due to the lower
flow velocity through the construction between the hulls. 3.5.4. Total drag coefficient of towed B1
The drag results are evaluated in terms of the total drag
coefficient (CD), the pressure drag coefficient (CPx) and the skin
3.5.3. Form factor friction drag coefficient (CF). Fig. 9 shows the numerical results of
For the length–diameter ratio 6:1 spheroid, estimates of (1 þk) the total drag coefficient (CD) at various transverse separations.
from Eq. (4) gives 1.135. The comparison of form factor between The CD of a single hull in the present study is 10% lower compared
the experimental results and the numerical results based on the to the experimental drag (Molland and Utama, 1997). A good
ITTC correction line is shown in Fig. 8. The numerically predicted agreement of the trend line is observed at S/L r0.37. At S/L Z0.47,
(1 þk) of B1 is within 1.5% error of Hoerner's estimation. The Molland and Utama (2002) suggested that some interference from
increment of form factor (1 þβk) compared with the single hull the tunnel wall may occur, and consequently, increase the total
illustrates the influence of viscous resistance factor (βk) due to drag of B1. This is not observed in either numerical study. The
close proximity. experimental result is 16% higher than the current numerical
The 8% error between the experimental result and Hoerner's prediction at S/L¼0.27 and 0.37. This may be partially attributed
estimation and 16% error between the experimental result and to interference drag due to the experimental mounting system,
numerical result show some interference of the tunnel wall to the which is not recreated in the numerical simulations. With these
single hull and twin hulls experiment, respectively. positive evaluations of the CFD with the experiment results and
30 P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35

Fig. 7. CSF as proportion of CD at various transverse separation.

Fig. 8. Comparison a form factor between the experimental and numerical


prediction of a single hull and twin hulls.

Fig. 6. Pressure drag coefficient of B1. (a) S/L ¼0.27, (b) S/L ¼ 0.37, and (c) S/L ¼0.47.

the mesh validation, this selected numerical method and the mesh Fig. 9. Comparison of CD of towed bare B1, experimental data made available at the
reference (Molland and Utama, 1997).
strategy has been proved to be effective and sufficient for the
further studies.
increase in the drag of individual members of the fleet. These
results may be classified into seven regions by the different sign of
4. Impact of spacing on drag %B1, %B2 and %CB as shown in Table 3 and are shown in Fig. 10. %B1
and %B2 is defined as the percentage difference of the individual
By varying the spacing of the fleet, the results suggest some drag of B1 and B2 referenced to a single hull drag, respectively. In
energy beneficial configurations exist for both individuals and the this paper, %CB is defined as the combined drag which considered
whole fleet. However, some configurations lead to a significant as a percentage difference of the fleet drag referenced to the sum
P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35 31

of two single hull drags. Where CD(B1) and CD(B2) is the drag 4.2. Echelon region
coefficient of B1 and B2 in fleet, respectively. CD(s) is the drag
coefficient of a single hull. With a transverse separation and a limited longitudinal offset
within one body length, the configuration is similar to echelon or part
%B1 ¼ fðC DðB1Þ  C DðsÞ Þ=C DðsÞ g  100; of a vee formation. In Fig. 13a and c, the results suggest the change in
%B2 ¼ fðC DðB2Þ  C DðsÞ Þ=C DðsÞ g  100; total drag is dominated by the pressure drag and the skin friction drag
at the close proximity at S/L¼0.17 and D/L¼ 0.27. Fig. 13a shows that
%CB ¼ fðC DðB1Þ þ C DðB2Þ  2C DðsÞ Þ=2C DðsÞ g  100 ð9Þ while both hulls experience interaction, the follower B2 experiences a
drag reduction due to the change in its pressure recovery profile and
The individual change in the total drag, the skin friction drag skin friction profile on the surface of B2 at40.6L, the leader B1
and the pressure drag for twin case compare with a single case are experiences a drag augment due to the change of the pressure profile
illustrated in Fig. 11a, b and c, respectively. and skin friction profile. The results show a similar magnitude of B1's
total drag augment and B2's total drag reduction (Fig. 10). The bigger
the longitudinal offset, the smaller the total drag increase.
4.1. Parallel region
4.3. No gain region
The parallel region is where both hulls are parallel to the
inflow, noses aligned with zero longitudinal offset (D/L¼ 0), both The individual drag of B1 and B2 is similar to the echelon
hulls experience a drag increment. In an infinite domain, both region; however, the combined drag in this region shows the
hulls are expected to experience an equal drag increase, however, minimal benefit of the fleet.
due to the asymmetric finite numerical fluid domain, the tunnel
walls result in slightly differing interference drag on each hull. 4.4. Push region
Fig. 12a–d illustrates the longitudinal pressure and skin friction
coefficient along the hull. It can be seen that the drag increase The push region is where the nose of B2 is positioned close to
experience is due to an increased skin friction drag rather than a the trailing edge of B1. The combined drag can be reduced by
change in pressure drag. The change in skin friction drag is due to 1–7%; B1 experiences a moderate reduction in drag due to the
an accelerated flow regime between the hulls. As the transverse presence of B2. There is insignificant change in skin friction
separation increases up to 0.47L, the accelerated flow reduces (Fig. 14c) and pressure distribution along the main body of both
towards the free-stream velocity, resulting in less than a 2% drag hulls (Fig. 14a), the variation in drag is dominated by the pressure
increase. These results and the previous studies (Hoerner, 1965; drag at the stagnation points (Rattanasiri et al., 2012). The pressure
Hucho and Ahmed, 1998; Molland and Utama, 2002) suggested distribution at the bow stagnation point of B2 reinforces the high
that the distance of 0.5L is required as the minimum transverse pressure at the stern of B1 leading to an increased pressure
separation to preclude body-to-body interaction. recovery. The results show the same effects as cars in convoy at
D/L ¼1.40L (Hucho and Ahmed, 1998). For example, at D/L E1.17L,
Table 3 the combined drag of twin spheroids is reduced by 7% while for
The body-to-body interaction region classified by spacings and drags. The cars in convoy are reduced by 15%. The individual drag of the
sign þand  indicates increasing and decreasing of the drag, respectively. leader is decreased by 20% and 30% for the spheroid and the car,
Name S/L D/L %B1 %B2 %CB
respectively, whilst the follower is increased by 7% and 20% for the
spheroid and the car, respectively. At S/L¼0, the drag of B1 is
Parallel region 0.17–0.47 0 þ þ þ reduced by 20% whilst drag of B2 is increased by 7%. Since the cars
Echelon region 0.17–0.47 0.27–0.57 þ  þ in the study of Hucho and Ahmed (1998) are less streamlines and
No gain region 0.37–0.47 0.87 þ  
therefore have a higher form factor (i.e. pressure drag), they
Push region 0.17 0.87–1.17  þ þ
0 o 1.37 experience more significant overall drag changes.
Drafting region 0 4 1.37   
Low interaction region 0.17 Z 1.47  þ  4.5. Drafting region
0.27 Z 0.87
0.37–0.47 1.17–1.47
No interaction region 0.37–0.47 Z 1.77 E0 E0 E0
By placing B2 directly behind B1 for D/L41.37, the combined
drag can be reduced by 5–6% with individual drag reduction

Fig. 10. The effect of twin hulls' spacing on individual drags and combined drag of B1 and B2 at Re¼ 3.2  106.
32 P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35

by 7–8% for B2 and 2–4% for B1. Considering Fig. 11a–c, as the
longitudinal offset increases, the results suggest a loss of energy
benefit. For D/L Z5, the drag reduction for B2 tends towards the
free stream drag value as wake recovery occurs (Rattanasiri et al.,
2012). This hydrodynamic behaviour is similar to cars in a convoy
with a slipstream at S/L 41.5 (Hucho and Ahmed, 1998). This goes
some-way to explain the energy benefit of cycling and swimming
in the slipstream (Kyle, 1979; Silva et al., 2008).

4.6. Low interaction region

Outside the previously proposed regions, the boundary of the


low interaction region is constructed, the detail of longitudinal
offset and transverse separation configurations are shown in
Table 3. The results in Fig. 10 show a minimal change of the
combined drag, less than 1% of drag reduction. It also suggests that
the larger the spacing between two hulls, results in less force
interaction. Individually, B2 gets a small drag increase whilst B1
gains some benefit. However, on average, the individual drag
exchange is less than 2%.

4.7. No interaction region

The results in Fig. 10 show almost no change of the individual


drag values and the combined drag due to both hulls experience
the free stream drag value as wake recovery occurs. The dominant
studies of spacing between twin hulls shows that the echelon
region, the no gain region, the low interaction region and the no
interaction region could potentially be used to lower propulsion
costs of an individual in the fleet.

5. Impact of configuration's shape on drag

Following the two hull study in the previous section, three and
four hulls configuration will now be considered. With the available
space of the domain, the three hulls in symmetry configuration
and the three/four hulls in echelon configuration can be studied to
determine the effect of the configuration's shape on drag. As
before %B3 and %B4 is defined as the percentage difference of
the individual drag of B3 and B4 referenced to a single hull drag,
respectively. Where CD(B3) and CD(B4) is the drag coefficient of B3
and B4 in fleet, respectively. The same consideration is applied to
the combined drag (%CB) for three and four hulls.
%B3 ¼ fðC DðB3Þ  C DðsÞ Þ=C DðsÞ g  100
: ð10Þ
%B4 ¼ fðC DðB4Þ  C DðsÞ Þ=C DðsÞ g  100

5.1. Impact of vee formation


Fig. 11. Effect of body-to-body interaction on forces acting on the drafting
configuration B1 and B2 at various longitudinal offsets. (a) Total drag coefficient,
By placing the two spheroids, B2 and B3, symmetrically along
(b) pressure drag coefficient, and (c) skin friction drag coefficient.
the centre line of the leading B1, the results show that the drag of
both following hulls (%B2 and %B3) is approximately the same.
Similar to the study of twin hulls, the effect of spacing among
three hulls can be classified into the same regions at the same The individual results show the similar number between the
specific spacings. The comparison of the drag of twin hulls and the leading member's drag reduction and the last member's drag
three hulls in vee formation (Table 4) shows that, with similar increase. Physically, 100% of individual drag reduction means hull
results of both B2 in the three arrangements, the drag increase of experiences the zero drag. For four hulls with a configuration of
B1 is twice as much as that of B1 in the twin formation. S/L ¼0.27 and D/L ¼0.27, the drag reduction of the last B4 is 129.3%,
the first 100% of drag reduction means that all of B4's hull
5.2. Impact of echelon formation resistance is eliminated, the rest of the drag reduction (29.3%)
illustrates the additional force from those leading hulls which
In Table 5, similar results to the twin hulls at S/L¼ 0.27 and resulting in pulling B4 forward. For the three and four bodies with
D/L¼ 0.27 and 0.57 configuration, show that the combined drag the configuration of S/L¼ 0.27 and D/L ¼0.57, both middle hulls
of three and four hulls has no energy benefit for the entire fleet. (B2) experience an approximately zero drag.
P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35 33

Fig. 12. Parallel configuration at D/L ¼ 0: the skin friction drag coefficient (CF) and the pressure coefficient (Cp). Where black line is a result of B1 as a single hull. The red dash
and blue dot lines are results of B1 and B2 in twin cases, respectively. (a) Cp at S/L ¼ 0.17, (b) Cp at S/L ¼0.37, (c) CF at S/L ¼ 0.17, and (d) CF at S/L ¼ 0.37. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 13. Echelon configuration at S/L ¼0.17 and 0.37 for fixed D/L ¼0.27: the skin friction drag coefficient (CF) and the pressure coefficient (Cp). Where black line is a result of
B1 as a single hull. The red dash and blue dot lines are results of B1 and B2 in twin cases, respectively. (a) Cp at S/L ¼ 0.17 and D/L ¼ 0.27, (b) Cp at S/L ¼0.37 and D/L ¼ 0.27, (c)
CF at S/L ¼ 0.17 and D/L ¼ 0.27, and (d) CF at S/L ¼ 0.37 and D/L ¼ 0.27. (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)

6. Conclusion drag and secondly, the influence of fleet configuration with multi-
ple hulls in the vee formation and the echelon formation.
The purpose of this paper was to determine firstly, the hydro- The dominant results for twin hulls' configurations show that
dynamic effect of the viscous interaction between two hulls on the spacing between hulls determines the drag of the entire fleet.
34 P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35

Fig. 14. Drafting configuration at S/L ¼ 0: the skin friction drag coefficient (CF) and the pressure coefficient (Cp). Where black line is a result of B1 as a single hull. The red dash
and blue dot lines are results of B1 and B2 in twin cases, respectively. (a) Cp at D/L ¼ 1.17, (b) Cp at D/L ¼ 1.77, (c) CF at D/L ¼ 1.17, and (d) CF at D/L ¼1.77. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 5
Table 4
Comparison of drags for the configuration of twin, three and four prolate spheroids
Comparison of drags for the configuration of twin, three spheroids in vee formation
in echelon at S/L ¼ 0.17 and D/L ¼ 0.27 and 0.57.
at S/L ¼ 0.17, 0.27 and 0.37 and D/L ¼0.27, 0.57, 0.87 and 1.17. Where %B2 and %B3 are
approximately the same.
D/L 0.27 0.57
S/L D/L 0.27 0.57 0.87 1.17
Twin Three Four Twin Three Four
Twin Sym Twin Sym Twin Sym Twin Sym
%B1 107.3 135.6 145.2 64 63.2 63.8
three three three three
%B2  93.9 12.1 43.6  62.4 0.6 0.4
%B3  199.6  13.9  60.2 3.4
0.17 %B1 107.3 217.4 64.0 126.8  15.8  38.9  11.9  24.2
%B4  129.3  61.7
%B2  93.9  92.1  62.4  59.1 12.8 19.4 10.3 14.1
%CB 6.7 11 0.8 2.8  1.5 0.0  0.8 1.4 %CB 6.7 9.5 11.4 0.8 1.3 1.4
0.27 %B1 55.6 112.8 38.1 76.6  2.2  5.7  6.4  13.0
%B2  49.7  48.3  37.3  35.5 0.7 3.1 5.4 7.3
%CB 3.0 5.3 0.4 1.8  0.7 0.2  0.5 0.6
The multi-vehicle configurations show limited influence over
0.37 %B1 33 68.7 26.0 54.6 3.5 7.3  2.5  5.4 the entire fleet's propulsive energy budget. However, for applica-
%B2  28.7  28.2  25.1  24.6  4.1  3.0 1.7 3.3
%CB 2.2 3.9 0.5 1.6  0.3 0.4  0.4 0.4
tions where lower propulsion cost for an individual is required, the
three ‘vee’ formation in the low interaction region could be
considered. Based on this numerical information, operators can
determine the optimal fleet configuration in transverse separation
Increasing the spacing results in lower the interaction. The and longitudinal offset based on energy considerations. The
dominant spacing has been characterised into seven zones based propeller's wake region is excluded from this simulation which
on the drag's characteristic of twin towed models. These are the by re-energising the wake region will impact on the drafting
parallel region, echelon region, no gain region, push region, vehicle. This is a topic for further study.
drafting region, low interaction region and finally the no interac-
tion region. The parallel region results in a drag increase for both
hulls with the smaller the spacing, the larger the drag increase
caused by viscous interactions. The echelon configuration could Acknowledgement
potentially be used to lower propulsion costs of an individual in
the fleet at the expense of the leading hull. Applications where The authors acknowledge the use of the IRIDIS High Perfor-
close proximity is required means that, the low interaction region mance Computing Facility and associated support services at the
can be indicated as the minimum spacing for minimum drag University of Southampton, in the completion of this work. The
increase. In these simulations, the drafting region provides the Ph.D. studentship of P. Rattanasiri was financed by the Royal
optimum configuration to minimise the overall fleet drag. Thai Government.
P. Rattanasiri et al. / Ocean Engineering 80 (2014) 25–35 35

References Larsson, L., Baba, E., 1996. Advances in Marine Hydrodynamics, vol. 5. Computa-
tional Mechanics Publications, Ohkusu, southampton, pp. 1–75 (Chapter: Ship
resistance and flow computations).
Aguiar, A.P., Almeida, J., Bayat, M., Cardeira, B., Cunha, R., Hausler, A., Maurya, P.,
Martins, A., Almeida, J.M., Silva, E., 2003. Coordinated maneuver for gradient search
Oliveira, A., Pascoal, A., Pereira, A., Rufino, M., Sebastiao, L., Silvestre, C., Vanni,
using multiple AUVs. In: Proceedings of OCEANS, vol. 1, pp. 247–352.
F., 2011. Cooperative control of multiple marine vehicles: theoretical challenges
Menter, F.R., 1994. Two-equation eddy-viscosity turbulence models for engineering
and practical issues. In: Proceedings of the 19th Mediterranean Conference on
applications. Am. Inst. Aeronaut. Astronaut. J. 32 (8), 1598–1605.
Control & Automation (MED). pp. 424–429.
Molland, A.F., Utama, I.K.A.P., 1997. Wind Tunnel Investigation of a Pair of Ellipsoids
Alexander, R.M., 2004. Hitching a lift hydrodynamically – in swimming, flying and
in Close Proximity. Technical Report 98. Ship Science, University of
cycling. J. Biol. 3. http://dx.doi.org/10.1186/jbiol5.
Andersson, M., Wallander, J., 2003. Kin selection and reciprocity in flight forma- Southampton.
tion? Behav. Ecol. 15, 158–162. Molland, A.F., Utama, I.K.A.P., 2002. Experimental and numerical investigations of a
ANSYS, 2010. ANSYS CFX, Release 12.1. ANSYS, Pennsylvania. pair of ellipsoids in close proximity. Proc. Inst. Mech. Eng., Part M: J. Eng. Marit.
Bean, T., Beidler, G., Canning, J., Odell, D., Wall, R., ORourke, M., Edwards, D., Environ. 216 (2), 107–115.
Anderson, M., 2007. AUV cooperative operations using acoustic communication Parsons, J.S., 1972. The Optimum Shaping of Axisymmetric Bodies for Minimum
and navigation. IEEE J. Ocean. Eng. Drag in Incompressible Flow (Ph.D. thesis). School of Mechanical Engineering,
Bellingham, J.G., Willcox, J.S., 1996. Optimizing AUV oceanographic surveys. In: Purdue University, Lafayette, IN.
Proceedings of the 1996 Symposium on Autonomous Underwater Vehicle Parsons, J.S., Goodson, R.E., Goldschmied, F.R., 1974. Shaping of axisymmetric bodies
Technology, AUV '96. 2–6 June 1996. for minimum drag in incompressible flow. J. Hydronaut. 8, 100–107.
Botelho, S., Neves, R., Taddei, L., 2005. Localization of a fleet of AUVs using visual Partridge, B.L., Johansson, J., Kalish, J., 1983. The structure of schools of giant bluefin
maps. In: Proceedings of OCEANS 2005 – Europe, vol. 2, pp. 1320–1325. tuna in cape cad bay. Environ. Biol. Fishes 9, 253–262.
Bradley, A.M., 1992. Low power navigation and control for long range autonomous Phillips, A., Haroutunian, M., Man, S., Murphy, A., Boyd, S., Blake, J., Griffiths, G.,
underwater vehicles. In: Proceeding of the Second International Offshore and 2012. Further Advances in Unmanned Marine Vehicles. In: Roberts, G.N., Sutton,
Polar Engineering Conference. San Francisco, CA. R. (Eds.), The Institution of Engineering and Technology, Stevenage, GB,
Bradley, A.M., Feezor, M.D., Singh, H., Sorrell, F.Y., 2001. Power systems for pp. 373–405 (Chapter: Nature in engineering for monitoring the oceans:
autonomous underwater vehicles. IEEE J. Ocean. Eng. 26, 526–538. comparison of the energetic costs of marine animals and AUVs).
Brath, T.J., Jesperson, D.C., 1989. The Design and Application of Upwind Schemes on Phillips, A., Turnock, S., Furlong, M., 2008. Comparisons of CFD simulations and in-
Unstructured Meshes. AIAA Paper 89-0366. service data for the self propelled performance of an Autonomous Underwater
Burger, M., Pavlov, A., Pettersen, K., 2009. Conditional integrators for path following Vehicle. In: Proceedings of the 27th Symposium of Naval Hydrodynamics.
and formation control of marine vessels under constant disturbances. In: Seoul, Korea, 05–10 October 2008.
Proceedings of the 8th IFAC International Conference on Manoeuvring and Phillips, A.B., 2009. Cost Effective Hydrodynamic Concept Design of Autonomous
Control of Marine Craft. pp. 179–184.
Underwater Vehicle (Ph.D. Thesis). School of Engineering Sciences, University
Cui, R., Ge, S.S., How, B.V.E., Choo, Y.S., 2009. Leader–follower formation control of
of Southampton, uk.
underactuated AUVs with leader position measurement. In: Proceedings of the
Phillips, A.B., Furlong, M., Turnock, S.R., 2007. The use of computational fluid
2009 IEEE International Conference on Robotics and Automation. Kobe, Japan,
dynamics to assess the hull resistance of concept autonomous underwater
pp. 2441–2446.
vehicles. In: Proceedings of OCEANS 2007 – Europe. Institute of Electrical and
Cui, R., Ge, S.S., How, B.V.E., Choo, Y.S., 2010. Leader–follower formation control of
underactuated AUVs. Ocean Eng. 37, 1491–1502. Electronics Engineers. Richardson, TX, USA.
Dalton, C., Zedan, M.F., 1980. Design of low-drag axisymmetric shapes by the Phillips, A.B., Turnock, S.R., Furlong, M., 2010a. Influence of turbulence closure
inverse method. J. Hydronaut. 15, 48–54. models on the vortical flow field around a submarine body undergoing steady
Edwards, D., Bean, T., Odell, D., Anderson,M., 2004. A leader–follower algorithm for drift. J. Mar. Sci. Technol. 15 (3), 201–217.
multiple AUV formations. In: Proceedings of IEEE/OES Autonomous Underwater Phillips, A.B., Turnock, S.R., Furlong, M., 2010b. The use of computational fluid
Vehicles. Sebasco, ME, vol. 1, pp. 40–46. dynamics to aid cost-effective hydrodynamic design of autonomous under-
Hanrahan, B., Juanes, F., 2001. Estimating the number of fish in atlantic bluefin tuna water vehicles. Proc. Inst. Mech. Eng., Part M: J. Eng. Marit. Environ. 224,
schools using models derived from captive school observations. Fish. Bull. 99 239–254..
(3), 420–431. Rattanasiri, P., Wilson, P.A., Phillips, A.B., 2012. Numerical investigation of the drag
Hoerner, S.F., 1965. Fluid-dynamic drag: practical information on aerodynamic drag of twin prolate spheroid hulls in various longitudinal and transverse config-
and hydrodynamic resistance (published by the author). urations. In: Proceedings of AUV 2012. Southampton, GB, 24–27 September
Hucho, W.H., Ahmed, S.R., 1998. Aerodynamics of Road Vehicles: From Fluid 2012.
Mechanics to Vehicle Engineering, 4th ed. Society of Automotive Engineers, Reeder, C.A., Odell, D.L., Okamoto, A., Anderson, M.J., Edwards, D.B., 2004. Two
Warrendale, PA. hydrophone heading and range sensor applied to formation-flying for AUVs. In:
Huggins, A., Packwood, A.R., 1994. The optimum dimensions for a long-range, Proceedings of OCEANS'04. MTTS/IEEE TECHNO-OCEAN '04, pp. 517–523.
autonomous, deep-driving, underwater vehicle for oceanographic research. Rhie, C.M., Chow, W.L., 1982. A Numerical Study of the Turbulent Flow Past an
Ocean Eng. 21, 45–56. Isolated Airfoil with Trailing Edge Separation. AIAA Paper 82-0998.
Husaini, M., Samad, Z., Arshad, M.R., 2009. CFD simulation cooperative AUV motion. Sarkar, T., Sayer, P.G., Fraser, S.M., 1997a. Flow simulation past axisymmetric bodies
Indian J. Mar. Sci. 38 (3), 346–351. using four different turbulence model. Appl. Math. Model. 21, 783–792.
ITTC, 1957. In: Proceedings of the 8th ITTC, Canal de Experiencias Hidrodinamicas. Sarkar, T., Sayer, P.G., Fraser, S.M., 1997b. A study of autonomous underwater
El Pardo, Madrid, Spain. vehicle hull forms using computational fluid dynamics. Int. J. Numer. Methods
Jagadeesh, P., Murali, K., 2005. Application of low-re turbulence models for flow
Fluids 25, 1301–1313.
simulations past underwater vehicle hull forms. J. Nav. Archit. Mar. Eng. 1 (2),
Silva, A.J., Rouboa, A., Moreira, A., Reis, V.M., Alves, F., Vilas-Boas, J.P., Marinho, D.A.,
41–54.
2008. Analysis of drafting effects in swimming using computational fluid
Jagadeesh, P., Murali, K., 2006. Investigation of alternative turbulence closure
dynamics. J. Sports Sci. Med. 7, 60–66.
models for axisymmetric underwater hull forms. J. Ocean Technol. 1 (2), 37–57.
Stevenson, P., Furlong, M., Dormer, D., 2007. AUV shapes – combining the practical
Jagadeesh, P., Murali, M., Idichandy, V.G., 2009. Experimental investigation of
hydrodynamic force coefficients over AUV hull form. Ocean Eng. 36, 113–118. and hydrodynamic considerations. In: Proceedings of OCEANS 2007 – Europe.
Karim, M.M., Rahman, M.M., Alim, M.A, 2009. Computation of turbulent viscous Vanni, F.V., 2007. Coordinated Motion Control of Multiple. Autonomous Under
flow around submarine hull using unstructured grid. J. Ship Technol. 5 (1), Vehicle (Master's thesis). Technical University of Lisbon, Lisbon, Portugal.
38–52. Weihs, D., 2004. The hydrodynamics of dolphin drafting. J. Biol. 3, 8.
Kinsey, J.C., 1998. Drag characterization in the autonomous benthic explorer. In: Weihs, D., Ringel, M., Victor, M., 2007. Aerodynamic interactions between adjacent
OCEANS'98 Conference Proceedings. 28 September–1 October 1998, pp. 1696– slender bodies instrumentation and wind tunnel testing. In: Proceedings of the
1700. 22nd International Congress on Instrumentation in Aerospace Simulation
Kyle, C.R., 1979. Reduction of wind resistance and power output of recing cyclists Facilities, ICIASF, pp. 1–28.
and runners traveling in group. Ergonomics 22, 387–397. Weiss, P., Mascarelli, J.V.C., Grossnet, D., Brignone, L., Labbe, D.F.L., Wilson, P.A.,
Labbe, D.F.L., Wilson, P.A., Weiss, P., Lapierre, L., 2004. Freesub: navigation guidance 2003. Freesub: dynamic stabilization and docking for autonomous underwater
and control of an IAUV. R. Inst. Nav. Archit. Trans. A: Int. J. Marit. Eng. 146 (2), vehicles. In: Proceedings of the 13th International Symposium on Unmanned
71–79. Untethered Submersible Technology. Durham, USA.

S-ar putea să vă placă și